#452452nd of 455 in Skills, plugins and prompts
osgrep
Local semantic code search CLI with call-graph tracing for coding agents
- Stars
- 1.2k
- License
- Apache-2.0
- Last commit
- Jan 2026
- Last release
- Dec 2025
- Language
- TypeScript
1,111 on 22 Apr 20261,150 on 11 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
osgrep is a grep-like CLI that indexes a repository with tree-sitter chunking, embeds it locally via onnxruntime-node, stores vectors in LanceDB and answers natural-language queries. It adds `trace` for callers and callees, `skeleton` for signature-only file views, and an optional `serve` daemon with file watching. Installers for Claude Code and opencode register hooks that start the daemon automatically.
Who it is for: Developers giving Claude Code or opencode local semantic code search
Strengths
- Embeddings run locally; models are about 150MB and no API key is needed
- Separate index per repository, keyed by git remote URL or directory hash
- `serve` keeps the index hot and re-indexes on file change; README claims <50ms responses
- Respects .gitignore and .osgrepignore; one-command install for Claude Code and opencode
Weaknesses
- Files without a tree-sitter grammar are skipped unless `--allow-fallback` is set
- Token and speed gains (20% and 30%) come from the author's own benchmark
- First search triggers a full index of the repo, which can be slow on large codebases
- Plugin integrations cover only Claude Code and opencode
What it needs
- Needs Node.js, npm
- Models: Local ONNX embedding models via onnxruntime-node, ColBERT reranking
- port 4444
What changed
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Also in Skills, plugins and prompts
See all 455| Rank | Project | Score |
|---|---|---|
| 1 | claude-memPersistent cross-session memory plugin for Claude Code and other coding agents | 97 out of 100 |
| 2 | impeccableDesign skill with 24 slash commands for AI coding agents' frontend output | 97 out of 100 |
| 3 | gstackMarkdown slash-command skills that give Claude Code planning, review, QA and ship roles | 96 out of 100 |
| #4 | rufloAgent harness adding swarms, memory and hooks to Claude Code and Codex | 96 out of 100 |
| #5 | cavemanShortens coding-agent replies and compresses tool output before the model reads it | 95 out of 100 |